Hi,
In a dataset with repeated measurements from each subject, I matched the exposure group to those who are unexposed to a certain factor. I have the data structure as follows:
Obs id Y X1 X2 matching
1 1 80 100 25 1
2 1 70 90 20 2
3 2 100 120 25 3
4 2 125 80 40 1
5 3 108 86 92 2
6 3 110 90 84 3
...
PROC MIXED DATA = aa METHOD = ML;
CLASS id;
MODEL Y = X1 X2;
REPEATED INT / TYPE = cs SUBJECT = id;
RUN;
In a matched study, the matching pairs (i.e., variable "matching" here) should be adjusted for to obtain unbias estimates. How can I adjust for matching? Treat matching as a binary variable and put it after X2, or in the REPEATED statement or allow it a RANDOM effect?
Thanks a lot?
Just so I get this straight, observations with the same value of matching form the pair, correct? I don't follow from the example data, where it looks like subject 1 is matched with subject 2 (obs 1 and 4) and subject 1 is also matched with subject 3 (obs 2 and 5). So pairings are not exclusive per subject, which strikes me as a bit odd. If we can get my head wrapped around the actual matching procedure, then I think the code will follow quickly, as the responses will be repeated measures on the pair.
Steve Denham
Yes. You are right. Pairs are not exclusive per subjects.
Ick.
I guess you could consider pair as a random effect, rather than repeated and try something like the following, where matching captures all of the pairs:
PROC MIXED DATA = aa METHOD = ML;
CLASS id matching;;
MODEL Y = X1 X2;
random matching;
REPEATED INT / TYPE = cs SUBJECT = id;
RUN;
Steve Denham
Thanks. I agree with you.
Are you ready for the spotlight? We're accepting content ideas for SAS Innovate 2025 to be held May 6-9 in Orlando, FL. The call is open until September 25. Read more here about why you should contribute and what is in it for you!
Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.